cifar10_h

  • תיאור :

גרסה מתויגת מחדש של ערכת הבדיקות של CIFAR-10 עם תוויות רכות המגיעות מכותבים אנושיים אמיתיים. עבור כל זוג (תמונה, תווית) בערכת הבדיקות המקורית של CIFAR-10, הוא מספק מספר תוויות נוספות שניתנו על ידי כותבים אנושיים אמיתיים, כמו גם התווית הרכה הממוצעת. מערך ההדרכה זהה לזה של מערך הנתונים המקורי.

לְפַצֵל דוגמאות
'test' 10,000
'train' 50,000
  • מבנה תכונה :
FeaturesDict({
    'annotator_ids': Sequence(Scalar(shape=(), dtype=int32)),
    'human_labels': Sequence(ClassLabel(shape=(), dtype=int64, num_classes=10)),
    'id': Text(shape=(), dtype=string),
    'image': Image(shape=(32, 32, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=10),
    'reaction_times': Sequence(Scalar(shape=(), dtype=float32)),
    'soft_label': Tensor(shape=(10,), dtype=float32),
    'trial_indices': Sequence(Scalar(shape=(), dtype=int32)),
})
  • תיעוד תכונה :
תכונה מעמד צוּרָה Dtype תיאור
FeaturesDict
annotator_ID רצף (סקלארי) (אף אחד,) int32
תוויות_אדם Sequence (ClassLabel) (אף אחד,) int64
תְעוּדַת זֶהוּת טֶקסט חוּט
תמונה תמונה (32, 32, 3) uint8
תווית ClassLabel int64
זמני_תגובה רצף (סקלארי) (אף אחד,) לצוף32
soft_label מוֹתֵחַ (10,) לצוף32
מדדי_ניסיון רצף (סקלארי) (אף אחד,) int32

רְאִיָה

  • ציטוט :
@inproceedings{wei2022learning,
  title={Human uncertainty makes classification more robust},
  author={Joshua C. Peterson and Ruairidh M. Battleday and Thomas L. Griffiths
  and Olga Russakovsky},
  booktitle={IEEE International Conference on Computer Vision and Pattern
  Recognition (CVPR)},
  year={2019}
}